A binomial modeling approach for upscaling colloid transport under unfavorable conditions: Emergent prediction of extended tailing

Abstract

AbstractColloid transport in saturated porous media is significantly influenced by colloidal interactions with grain surfaces. Near‐surface fluid domain colloids experience relatively low fluid drag and relatively strong colloidal forces that slow their downgradient translation relative to colloids in bulk fluid. Near‐surface fluid domain colloids may reenter into the bulk fluid via diffusion (nanoparticles) or expulsion at rear flow stagnation zones, they may immobilize (attach) via primary minimum interactions, or they may move along a grain‐to‐grain contact to the near‐surface fluid domain of an adjacent grain. We introduce a simple model that accounts for all possible permutations of mass transfer within a dual pore and grain network. The primary phenomena thereby represented in the model are mass transfer of colloids between the bulk and near‐surface fluid domains and immobilization. Colloid movement is described by a Markov chain, i.e., a sequence of trials in a 1‐D network of unit cells, which contain a pore and a grain. Using combinatorial analysis, which utilizes the binomial coefficient, we derive the residence time distribution, i.e., an inventory of the discrete colloid travel times through the network and of their probabilities to occur. To parameterize the network model, we performed mechanistic pore‐scale simulations in a single unit cell that determined the likelihoods and timescales associated with the above colloid mass transfer processes. We found that intergrain transport of colloids in the near‐surface fluid domain can cause extended tailing, which has traditionally been attributed to hydrodynamic dispersion emanating from flow tortuosity of solute trajectories.